Mini assignment 1
Kedar Patil
February 16, 2015
NYC Motor Vehicle Collision Data
Background
This data has been downloaded from as of 02/15/2015, NYC Open Data website. It has 531,574 records and covers activity from 07/01/2012 to 02/10/2015. This analysis focuses only on full year 2013-2014.
Preliminary Findings
Strong seasonality is observed, with injuries peaking around beginning of summer, when more people are out on street.
Motor vehicle injuries form the biggest share of injuries, followed by pedestrians. Number of injuries has not changed substantially across two years.
Further, pedestrians and cyclists have somewhat opposite relationship when it comes to injuries
Passenger and SUVs contribute most to injuries
Looking for intersections that have had >= 15 pedestrians injured in last 2 years, we see clusters in mid-town and upper manahattan and Bronx. There are a few spots in Queens and Brooklyn as well.
Major streets and intersections around 14th, 23rd, 34th, 42nd, and Times Square, as well as 119th, 125th, and 135th streets seem to have high number of injuries.
Code Details:
R script used, along with resulting charts, are given below
nycmvc_data <- read.csv("/Users/kedarpatil/edav/data/NYPD_Motor_Vehicle_Collisions.csv")
#FIXING DATES
nycmvc_data$DATE <- as.Date(nycmvc_data$DATE, format='%m/%d/%Y')
#CREATING SUMMARY TABLE
library(plyr)
mth_frq_all <- ddply(nycmvc_data, c(year_grp="strftime(nycmvc_data$DATE, format='%Y')",
mnth_grp="strftime(nycmvc_data$DATE, format='%m')"),
summarise, Inj_Pers=sum(NUMBER.OF.PERSONS.INJURED),
Fat_Pers=sum(NUMBER.OF.PEDESTRIANS.KILLED))
#CHARACTER TO NUMERIC
mth_frq_all$mnth_grp <- as.numeric(mth_frq_all$mnth_grp)
#PLOT 2013-2014 INJURIES ONLY
library(dplyr)
fltr_mth_frq_all <- filter(mth_frq_all, year_grp=="2013"|year_grp=="2014")
#PLOT INJURIES
require(ggplot2)
require(scales) # for removing scientific notation
(ggplot(fltr_mth_frq_all, aes(x=mnth_grp, y=Inj_Pers, color=year_grp, xmin=1, xmax=12, ymin=0, ymax = 7000))
+ geom_line(size=1) + geom_point(size=3)
+ ggtitle("YoY Overall Injuries - 2013-2014")
+ scale_y_continuous(labels = comma)
+ scale_x_discrete(labels = comma)
+ theme(plot.title=element_text(size=rel(1.5), face="bold"))
+ theme(axis.title=element_text(size=15),
axis.text=element_text(size=10),
legend.text=element_text(size=10),
legend.title=(element_text(size=10)))
+ labs(x="Month", y="Observed Frequency", color="Year"))
#PLOT 2013-2014 INJURIES ONLY - DEEPER DIVE
mth_frq_deep <- ddply(nycmvc_data, c(year_grp="strftime(nycmvc_data$DATE, format='%Y')",
mnth_grp="strftime(nycmvc_data$DATE, format='%m')"),
summarise,
Pedestrians=sum(NUMBER.OF.PEDESTRIANS.INJURED),
Cyclists=sum(NUMBER.OF.CYCLIST.INJURED),
Motorists=sum(NUMBER.OF.MOTORIST.INJURED))
#CHARACTER TO NUMERIC
mth_frq_deep$mnth_grp <- as.numeric(mth_frq_deep$mnth_grp)
library(dplyr)
fltr_mth_frq_deep <- filter(mth_frq_deep, year_grp=="2013"| year_grp=="2014")
#MELTING DATA
require(reshape2)
tmp_melt <- melt(fltr_mth_frq_deep,
id = c("year_grp", "mnth_grp"),
variable.name = "Injury_Category", value.name = "Freq")
tmp_melt_bar <- ddply(tmp_melt, c("year_grp","Injury_Category"),
summarise, "Frequency" = sum(Freq))
require(scales) # for removing scientific notation
(ggplot(tmp_melt_bar, aes(x=Injury_Category, y=Frequency, fill=year_grp, ymax = 45000)) + geom_bar(position="dodge",stat="identity")
+ geom_text(aes(label=comma(Frequency)), vjust=-1, size=3.5, position=position_dodge(.9))
+ scale_y_continuous(labels = comma)
+ ggtitle("YoY Injuries by Category - 2013-2014")
+ theme(plot.title=element_text(size=rel(1.5), face="bold"))
+ theme(axis.title=element_text(size=15),
axis.text=element_text(size=10),
legend.text=element_text(size=10),
legend.title=(element_text(size=10)))
+ labs(x="Injury Category", y="Observed Frequency", fill="Year"))
#PLOT 2013-2014 INJURIES ONLY - DEEPER DIVE - LINE
tmp_melt_line <- ddply(tmp_melt, c("mnth_grp","Injury_Category"),
summarise, "Frequency" = sum(Freq))
(ggplot(tmp_melt_line, aes(x=mnth_grp, y=Frequency, color=Injury_Category, xmin=1, xmax=12, ymin=0))
+ geom_line(size=1) + geom_point(size=3)
+ ggtitle("Injuries by Category - 2013-2014")
+ scale_y_continuous(labels = comma)
+ scale_x_discrete(labels = comma)
+ theme(plot.title=element_text(size=rel(1.5), face="bold"))
+ theme(axis.title=element_text(size=15),
axis.text=element_text(size=10),
legend.text=element_text(size=10),
legend.title=(element_text(size=10)))
+ labs(x="Month", y="Observed Frequency", color="Category"))
#PLOT VEHICLE TYPE AND INJURY - HORIZONTAL BAR CHART
inj_veh_typ <- ddply(filter(nycmvc_data, DATE>"2012-12-31" & DATE<"2015-01-01"),
c("VEHICLE.TYPE.CODE.1"),
summarise,
Pedestrians=sum(NUMBER.OF.PEDESTRIANS.INJURED),
Cyclists=sum(NUMBER.OF.CYCLIST.INJURED),
Motorists=sum(NUMBER.OF.MOTORIST.INJURED))
tmp_melt <- melt(inj_veh_typ,
id = c("VEHICLE.TYPE.CODE.1"),
variable.name = "Injury_Category", value.name = "Frequency")
tmp_melt_bar <- filter(tmp_melt, Injury_Category=="Pedestrians")
tmp_melt_bar <- filter(tmp_melt, Injury_Category=="Cyclists")
tmp_melt_bar <- filter(tmp_melt, Injury_Category=="Motorists")
require(scales) # for removing scientific notation
(ggplot(tmp_melt_bar, aes(x=reorder(VEHICLE.TYPE.CODE.1,Frequency), y=Frequency, fill=VEHICLE.TYPE.CODE.1))
+ geom_bar(stat="identity")
+ coord_flip()
#+ geom_text(aes(label=comma(Frequency)), vjust=-1, size=6, position=position_dodge(.9))
+ scale_y_continuous(labels = comma)
+ ggtitle("Injuries by Vehicle Type - 2013-2014 - Pedestrians")
# + ggtitle("Injuries by Vehicle Type - 2013-2014 - Cyclists")
# + ggtitle("Injuries by Vehicle Type - 2013-2014 - Motorists")
+ theme(plot.title=element_text(size=rel(1), face="bold"))
+ theme(axis.title=element_text(size=15),
axis.text=element_text(size=10),
# legend.text=element_text(size=10),
legend.title=(element_text(size=10)))
+ guides(fill=FALSE)
+ labs(x="Injury Category", y="Observed Frequency", fill="Vehicle"))
#CREATING MAPS
library(plyr)
injureped=subset(nycmvc_data,NUMBER.OF.PEDESTRIANS.INJURED > 0)
injureped_loc <- ddply(injureped, c("LONGITUDE", "LATITUDE"),
summarise, Inj_No=sum(NUMBER.OF.PERSONS.INJURED))
injureped_loc_sum <- ddply(injureped_loc, c("Inj_No"),
summarise, FreqCnt=length(Inj_No))
high_injure_loc_ped <- subset(injureped_loc, Inj_No >= 15)
library(ggmap)
nycmap <- get_map(location = "Queens, New York", zoom=11)
(nyc1 <- ggmap(nycmap) + scale_color_discrete(name = ">=15 Pedestrians Injured - 2013-2014")
+ ggtitle(">=15 Pedestrians Injured - 2013-2014")
+ theme(plot.title=element_text(size=rel(1), face="bold"))
+ theme(legend.text=element_text(size=0),
legend.title=(element_text(size=0)),
legend.position="none")
+ geom_point(data=high_injure_loc_ped, aes(LONGITUDE, LATITUDE, color = 'Pedestrians'), size=2, alpha=1))
#Midtown-downtown Map
nycmap <- get_map(location = "Flatiron, New York", zoom=13)
(nyc1 <- ggmap(nycmap) + scale_color_discrete(name = ">=15 Pedestrians Injured - 2013-2014")
+ ggtitle(">=15 Pedestrians Injured - Mid, Lower Manhattan")
+ theme(plot.title=element_text(size=rel(1), face="bold"))
+ theme(legend.text=element_text(size=0),
legend.title=(element_text(size=0)),
legend.position="none")
+ geom_point(data=high_injure_loc_ped, aes(LONGITUDE, LATITUDE, color = 'Pedestrians'), size=3))
#Uptown and Brox map
nycmap <- get_map(location = "Harlem, New York", zoom=14)
(nyc1 <- ggmap(nycmap) + scale_color_discrete(name = ">=15 Pedestrians Injured - 2013-2014")
+ ggtitle(">=15 Pedestrians Injured - Upper Manhattan & Bronx")
+ theme(plot.title=element_text(size=rel(1), face="bold"))
+ theme(legend.text=element_text(size=0),
legend.title=(element_text(size=0)),
legend.position="none")
+ geom_point(data=high_injure_loc_ped, aes(LONGITUDE, LATITUDE, color = 'Pedestrians'), size=3))